1. Field of the Invention
The present invention relates, in general, to a method and apparatus for creating improved inductors capable of being operated at relatively low voltage levels and further capable of being implemented on an integrated circuit chip. Specifically, the present invention relates to a method and apparatus for creating improved inductors, capable of being operated at relatively low voltage levels and further capable of being implemented on an integrated circuit chip, and which can be adapted for use with electronic oscillators and thereby provide improved Q factors for electronic oscillators.
2. Description of Related Art
An electronic oscillator is a device or circuit that produces a periodically varying output at a controlled frequency. The electronic oscillator's output can be either voltage, current, or an electromagnetic wave.
Electronic oscillators may be either passive or active. Passive electronic oscillators are electronic circuits composed of passive electrical components. Active electronic oscillators are electronic circuits composed of at least one active electrical component, and any number of passive electrical components. Passive electrical components are those components which cannot independently generate electrical energy. Active electrical components are those components which can independently generate electrical energy.
Ideally (i.e., mathematically), electronic oscillators generate electrical energy precisely at a predefined controlled frequency. Practically (i.e., in the actual physical world), electronic oscillators generate electrical energy concentrated around a predefined controlled frequency; that is, actual (as opposed to mathematical) electronic oscillators do not generate electrical energy precisely at a predefined controlled frequency, but rather tend to generate electrical energy within a "band" of frequencies spanning some predefined controlled frequency range. This is true irrespective of whether passive or active electronic oscillators are considered.
Generally, when either passive or active electronic oscillators are actually built, the objective is to make such actual electronic oscillators approximate, as nearly as possible, their ideal mathematical equivalents. That is, an attempt is made to (1) make the "band" of frequencies spanned by the actual oscillator as narrow as possible, and (2) center that "band" of frequencies as nearly as possible about the predefined controlled centerline frequency of the ideal mathematical electronic oscillators that the actual electronic oscillator is intended to emulate.
Insofar as the objective in building an actual passive or active electronic oscillator is to approximate the actual electronic oscillator's mathematical counterpart, the "quality" of such actual electronic oscillator is assessed relative to how closely such actual electronic oscillator approximates its ideal mathematical counterpart. The "center frequency" of an actual electronic oscillator is defined to be that frequency where the output electrical power of the actual electronic oscillator is at a maximum. The usable "bandwidth" of an actual electronic oscillator is defined to be that range of frequencies, centered about the defined "center frequency," where the output electrical power has dropped to one-half the value of the maximum output electrical power (i.e., the defined "center frequency"). The "quality" of an actual electronic oscillator is described quantitatively in terms of a "Quality Factor" (Q Factor) which is defined to be the ratio of the defined "center frequency" to the defined "bandwidth." In symbols, the Q factor is typically expressed as follows: Q=.omega..sub.0 /(.omega..sub.2 -.omega..sub.1), where (.omega..sub.0 stands for the "center frequency" of the oscillator, and (.omega..sub.2 -.omega..sub.1) stands for the "bandwidth" of the oscillator.
With respect to actual oscillator design, then, the objective is to build an oscillator with as high a "quality," or Q Factor, as is practicable, since the higher the Q Factor, the more closely the actual oscillator approximates a mathematically ideal oscillator. As defined, the Q Factor is a mathematical formula requiring both the defined "center frequency" and the defined "bandwidth" of an actual oscillator. In practice, the "center frequency," defined "bandwidth," and thus the Q Factor of actual oscillators are assessed by use of a spectrum analyzer.
A spectrum analyzer is a device which visually depicts the electrical power of a signal distributed over a range of frequencies. In the visual display portion of a typical spectrum analyzer, such distribution of electrical power in a signal is displayed by means of a graph. The horizontal axis of the graph is marked using units of frequency, and the vertical axis of the graph is marked using units of power per unit of frequency. Using this graph, the power of an actual oscillator's output can be plotted over a range of frequencies. Typically, the power versus frequency plot for an actual oscillator will appear as a "bell shaped" curve with the apex of the "bell shaped" curve corresponding to the "center frequency," as defined above, and with the "bandwidth" consisting of that range of frequencies within which is contained one-half (1/2) the total power within the output of the oscillator as is represented by the area under the curve.
The better an actual oscillator is, the narrower such oscillator's "bell shaped" curve will appear on a spectrum analyzer. This is true because a higher Q factor generally connotes a smaller defined "bandwidth," which indicates that the majority of electrical energy produced by the oscillator is concentrated about the predefined controlled frequency. With respect to a spectrum analyzer, such higher Q Factors translate to narrower "bell shaped" curves, since the "bandwidth" (frequency band which contains one-half (1/2) the total power contained within the output of the oscillator) is narrower for higher Q Factors.
In practice, it has been found that one of the most significant factors which broadens, or spreads, the "bandwidth" of an oscillator is the time-domain "jitter" of the oscillator. The mathematically ideal oscillator, described above, produces an output waveform of some predefined controlled frequency. The frequency of a waveform is defined to be one divided by the period of time that elapses between successive wave crests of the output waveform. In the mathematically ideal oscillator the period of time between any two successive wave crests in the mathematically ideal oscillator's output waveform is always the same. Unfortunately, this is not the case with actual oscillators.
In actual oscillators, the period of time between any two successive wave crests in an actual oscillator's output waveform varies. In the context of electronic oscillators, the term utilized to describe this phenomenon of abrupt variations in the periods of an oscillator's successive output waveforms is time-domain "jitter."
This "jitter" is responsible for the "spread" of the "bell shaped" curve of an actual oscillator, as such output appears on a spectrum analyzer. That is, because each such variation in period constitutes a wave of different frequency, the power output of the oscillator will appear as spread over the range of frequencies present when such output is viewed via spectrum analyzer. The more numerous the different frequencies produced by the oscillator (i.e., the higher the time-domain "jitter"), the larger the frequency range over which the output of the oscillator is spread, and, consequently, the wider the "bandwidth," and thus the lower the Q Factor of the oscillator. Thus, from the standpoint of oscillator design it is important that such time domain "jitter" be kept as low as practicable in order to provide the best Q Factor possible within design constraints.
As can be seen from the foregoing, both "Q Factor" and time-domain "jitter" are alternate and inverse ways of describing the accuracy of an actual oscillator; that is, a high Q factor implies relatively low time-domain "jitter," and vice versa.
The accuracy required of an actual oscillator is dictated by the application in which the oscillator will be used. One type of application which requires oscillators having significant accuracy is the adaptation of electronic oscillators for use as a clock in a data-processing system.
A clock in the data-processing system is a device that generates periodic, accurately spaced signals used for various purposes such as timing, synchronization, and regulation of the operations of a data processor within the data-processing system, or the generation of interrupts. Since electronic oscillators, by definition, produce periodic accurately spaced signals, it is common to adapt them for use as clocks in data-processing systems.
Both passive and active electronic oscillators are used as clocks in data-processing systems. For data-processing systems requiring clock speeds of 200 MHZ and below, it is common to use active oscillators to provide the clock signal. Typically, such active oscillators are in the form of what are known in the art as "ring oscillators." While there are different forms of "ring oscillators," the basic form of such oscillators is that of differential-type inverters which have gain and/or a total phase delay of a negative one-hundred-eighty (-180) degrees. As stated, such active oscillators tend to work well for clock speeds of 200 MHZ and below; however, for data-processing systems requiring clock speeds greater than 200 MHZ, such active oscillators are generally too inaccurate to provide such required clock speeds in that such active oscillators have unacceptably high levels of time-domain "jitter" and correspondingly low Q factors.
The primary sources of such unacceptably high levels of time-domain "jitter," and correspondingly low Q Factors, are inherent in the components utilized to construct such active oscillators. Under the current state of the art at least one of the primary noise sources cannot be eradicated in that it arises from thermal noise in the inverter transistors, which is a physical property inherent in the materials used to construct the inverter transistors. Furthermore, while a second primary noise source (time-domain "jitter" arising from variations in power supply voltage supplied to the active inverters) can be reduced by careful control of the power supply voltage, it cannot be eradicated and will in fact become a significant source noise at high frequencies (e.g., those frequencies starting in the 600-800 MHZ range and extending to the 1 GHz range and beyond). In addition to these two primary sources of noise, there are additional sources of noise which are also inherent in the components utilized to construct such active oscillators, such as substrate coupling in the oscillator when it is operating on a digital chip, or additional background coupling through the substrate ("ground bounce") which also cannot be eradicated due to the fact that such noise sources are inherent in the components utilized to construct such active oscillators.
Within the art, integrated circuit designs are trending toward lower and lower voltages; for example, integrated circuit designs are currently moving from 1.8 volts to 1.5 volts. This trend makes the previously-noted significant noise contributors even more significant. Present methods attempting to offset such noise contributions have included providing a special supply voltage of 2.4 volts to the oscillator itself. However, such methods begin to fail near the 600-800 MHZ range.
Attempts have been made within the art to develop other types of active oscillators which do not have the foregoing noted noise source problems. One such type of oscillator that has been developed is the surface acoustical wave (SAW) oscillator. The SAW-type oscillators tend to be very accurate, with very high Q factors. However, such SAW-type oscillators are not without practical problems. For example, (1) SAW-type oscillators generally require a two chip implementation, which can be very inconvenient in a data-processing system context; (2) the SAW-type oscillators are best tuned to frequencies within the 200-800 MHZ range, although some manufacturers are currently offering SAW-type oscillators up to 1.2 GHz; (3) SAW-type oscillators are very sensitive to temperature changes; (4) the frequency output of such SAW-type oscillators tends to vary with temperature; and (5) SAW-type oscillators typically have no tunability. Furthermore, SAW-type oscillators tend to be relatively expensive, especially in the very high frequency range.
In light of the foregoing, it is apparent that a need exists for a method and apparatus for creating an improved inductor, capable of being operated at relatively low voltage levels and further capable of being implemented on an integrated circuit chip, which can be adapted for use with an electronic oscillator and which can provide such an oscillator with a relatively high Q factor and correspondingly low time-domain "jitter."